New Supervised and Unsupervised Machine-Learning Methods on P38 Mapk Longevity Reveals Regulation of Age-Dependent Disease Proteins
Maintaining protein homeostasis is an important aspect of cellular health, however, this process becomes impaired with aging leading to cellular dysfunction. We have previously found that the stress response protein p38 MAPK (p38Kb) regulates lifespan and age-dependent locomotor behaviors in Drosophila, via a novel role in mediating protein homeostasis. We have developed new machine-learning approaches to perform proteomic analysis of muscle and brain tissues across the entire lifespan of the long-lived p38Kb over-expression animals, short-lived p38Kb mutants, and genetic background controls to determine the changes in the proteome that may be influencing longevity versus accelerated aging. Using unsupervised cluster analysis of differentially expressed proteins, we have discovered sets of proteins enriched with functional annotations. We also used cluster analysis to infer regulatory interactions between differentially expressed proteins. Using supervised machine learning to identify predictors of aging in control flies, we find that the brain and muscle age differently with unique suites of proteins driving the aging process. We also used our machine learning model to predict the age of p38K over-expression and mutant animal samples. We find that the long-lived p38K over-expression animals exhibit a younger profile. Conversely, loss of p38K leads to an accelerated shift with animals prematurely entering older profiles. Interestingly, we find that p38K regulates the levels of a subset of these muscle and brain aging predictor proteins, suggesting that p38K may coordinate a node that drives the normal rate of aging.
Becerra, Basheer, "New Supervised and Unsupervised Machine-Learning Methods on P38 Mapk Longevity Reveals Regulation of Age-Dependent Disease Proteins" (2018). University Research Symposium. 10.